1. Field of the Invention
This invention relates generally to optoelectronic systems for imaging objects from an elevated or slightly elevated observing instrument. Such imaging systems include but are not limited to mast-mounted systems for obtaining warning of shallow hazards ahead of a water craft, aircraft-carrier landing aids, and refinements in airborne imaging platforms. A related aspect of the invention provides intensity equalization across a fan-shaped probe beam, and has general industrial applications.
2. Related Art
Shallow-angle marine observation systems—A particular difficulty of all marine observational systems, even visual systems, is the problem of interference by the water surface. Reflections at the surface, whether of ambient radiation or of probe beams, tend to be confused with signals or signatures of the hazards or other objects of interest.
Another noteworthy problem with such systems is the limited range of known apparatus and methods. In the past, short range has been seen as essentially an inherent limitation of mast-mounted or other only-slightly-elevated equipment.
It is known to use light detection and ranging (“LIDAR”) for such purposes. FIG. 1 illustrates an experimental deployment shown by Anderson, Howarth and Mooradian (“Grazing Angle LIDAR for Detection of Shallow Submerged Objects”, Proc. International Conference on Lasers, 1978).
Anderson et al. did a pier-based experiment with a single-pixel PMT detector and no scanner. Basically they verified the laws of physics, namely (1) Snell's Law predicting deflection of the light into the water, and (2) the laws of radiative transfer—the light detection and ranging or “LIDAR” equation—predicting enough returning photons to support a detection. There was no suggestion of an entirely practical implementation for such an idea.
More specifically, the Anderson paper describes use of grazing-incidence LIDAR for detection of shallow objects. The group detected a target of diameter about 80 centimeters (2½ feet), to depths of nearly 5 meters (15 feet) at a range of 130 meters (400 feet) from a pier.
The experimental demonstration used a narrow-beam LIDAR and a photomultiplier-tube detector. The laser L (FIG. 1) and receiver R were mounted in a hut-like enclosure E on a pier structure S in the ocean, at distance F of about 330 m (1100 feet) forward from the beach.
The LIDAR transceiver L-R was at a height H of about 13 m (40 feet) above the ocean surface O. At the pier the benthic depth D1 was some 5 m (15 feet) and at the target T the depth D2 was about 8 m (25 feet).
A winch W on the pier operated a chain CH around a first pulley P1, fixed by a clamp CL to the pier S. The chain extended out to the floating target T via a second pulley P2, which was tethered to an anchor A (in the form of concrete-filled 55-gallon drums)—thus enabling some variation in range R as desired, the nominal value of the range R being 120 m (400 feet). The severely constrained range associated with these experiments is exemplary of the limitations of shallow-angle object surveillance heretofore.
We are aware of these patents for mast-mounted television cameras used for imaging objects from slightly elevated positions: U.S. Pat. Nos. 3,380,358 and 3,895,388. Pertinent LIDAR-related patents include:                U.S. Pat. No. 4,862,257 of Ulich,        U.S. Pat. No. 4,920,412 of Gerdt,        U.S. Pat. No. 5,013,917 of Ulich        U.S. Pat. No. 5,034,810 of Keeler,        U.S. Pat. No. 5,091,778 of Keeler,        U.S. Pat. No. 5,257,085 of Ulich,        U.S. Pat. No. 5,450,125 of Ulich,        U.S. Pat. No. 5,384,589 of Ulich and        U.S. Pat. No. 5,506,616 of Scheps.The most relevant of these are the last three Ulich patents mentioned.        
Ulich et al. use a streak tube for time-resolved fluorescence (wavelength vs. time), not imaging (angle vs. time). In fact, their text particularly cites use of a streak tube in a nonimaging mode. Furthermore they use a laser blocking filter to specifically reject the in-band response.
Thus the prior art fails to deal incisively, or effectively, with the previously mentioned problems of interference arising from surface reflection. Utilization of a slit by the Ulich group is for spectral dispersion, not imaging.
The '589 Ulich patent, “Imaging LIDAR System”, makes one reference to a ship-based application, but does not develop the idea further. The system is described only with reference to gated, intensified cameras.
Airborne-hazard alert for water craft—LIDAR is also usable for obtaining information about airborne objects, whether threatening hostile objects or otherwise. A separate system for such purposes, however, is costly and occupies significant space in the command center of a water craft.
Aircraft-carrier operations—In addition to detection of floating and airborne obstacles (e. g. mines and other hazards), another marine-related problem that would benefit from visibility aids is that of aircraft-carrier landing. This problem is particularly acute at night, and in fog or other turbid-atmosphere conditions.
The difficulty of such operations is compounded by the high speeds involved, the fact that not only the aircraft but also the carrier is in motion. A further complication sometimes is the need for a degree of discreet or covert character in the traffic. Radio guidance may be of limited practicality in such circumstances.
Airborne surveillance—Still another use of LIDAR systems that has been developed heretofore is airborne surveillance of objects submerged in the ocean or in other bodies of water. U.S. Pat. No. 5,467,122—commonly owned with the present document—sets forth many details of a surveillance system that is particularly aimed at monitoring relatively large areas of the ocean.
In that system, typically imaging is limited to detection from altitudes of at least 160 m (500 feet) and looking straight down into the water with the center of the probe beam. Still, there is some off-axis detection for positions well away from the track of the airborne platform.
Wave noise, and distortion: Wave noise and the resultant image distortion represent one of the severest limitations for airborne surveillance, even in the clearest ocean waters. These concerns have not been adequately addressed with existing airborne LIDAR systems. According to a comparative-evaluation field test in 1997, object-classification capability and the ability to reject false alarms in hazard detection have yet to be achieved to the satisfaction of the United States government.
Both the shapes and the positions of submerged objects are distorted by uncorrelated refractions of different parts of the probe/return beam, due to irregularity of the water surface. Heretofore no effort has been directed to overcoming either the positional error or the relative vagueness of object shapes obtained with this technology.
Uncertainties in coverage: Current systems also provide inadequate information about the fraction of the undersea environment that is actually being screened. The root problem is that wave focusing and defocusing of rays from a LIDAR system cause gaps in the coverage at different depths.
That is to say, inherently certain volumes of water receive and reflect very little light, which means that objects within those volumes cannot be detected. The difficulty here is that existing systems cannot accurately estimate the extent of these effects at different depths, and therefore cannot generate good area-coverage estimates at those depths.
There is no reliable measurement of how well—in particular, how uniformly—the system is illuminating and imaging each layer of water. Such systems resort to a statistical model, based on a single estimate of sea state, to estimate how many passes over the same patch of water are necessary to assure proper coverage.
This model is hard to validate—and the estimate of sea state may or may not be accurate or timely. Errors in the sea-state estimate force present systems to make either too many passes over the same area, which results in poor effective area-coverage rates, or too few passes, which may leave the area inadequately sampled and so unsafe for ship transit.
Refractive-correction: A hitherto unrelated technology is reported in another coowned patent, U.S. Pat. No. 5,528,493—which teaches use of observations from below an irregular water surface, i. e. by stationary or very slowly moving submerged apparatus. This latter patent refines images collected in such observations by correcting for effects of refraction at each point of the surface.
No effort heretofore has been directed to adapting this technology to either surveillance of the sea from either aircraft or surface water craft. This method requires a height map of the ocean-wave surface—to find all the refraction directions and so solve Snell's law for each spot.
To obtain such a height map, preferred forms of the patented method depend in turn upon iterative determination of the dynamic surface condition. These forms of the method therefore rely heavily upon the essentially stationary character of the observing platform, and are accordingly too slow for use with fixed-wing surveillance aircraft.
Since the method of the '493 patent is able to determine only bearing, not range—from any single observation apparatus—image reconstruction is impossible from such a single apparatus. Image reconstruction accordingly requires data from two observation subsystems separated by a known baseline and working in tandem.
Reconstruction is then accomplished through the sort of dual-station baseline triangulation that is familiar in surveying. Although the requirement of a long baseline is acceptable for waterborne observation platforms that are very large, and so intrinsically can provide a long baseline, such a requirement is undesirable for airborne surveillance as it calls for a very large aircraft operating at very low altitudes—or alternatively introduces the additional complications of plural aircraft conducting a coordinated surveillance.
Algorithms to reconstruct the distorted images of underwater targets as seen from above the surface have, however, been developed by M. S. Schmalz et al. Some of this work is reported in “Rectification of refractively-distorted imagery acquired through the sea surface—an image algebra formulation”, in Proceedings SPIE 1350 (1990); and “Errors inherent in the restoration of imagery acquired by viewing through remotely-sensed refractive interfaces and scattering media”, in Proceedings SPIE 1479 (1991).
In the reported work, a subsurface image is reconstructed iteratively, starting with assumptions about the depth of observed objects. Results from the Schmalz group have not been applied in the LIDAR context, or to airborne surveillance generally.
Glint interference with volume backscatter: Another hitherto unrelated field of work, previously addressed only in the context of bottom mapping, is due to G. C. Guenther et al. For decades they have studied and documented the problem of confusion between surface glints and probe-beam backscatter from the ocean volume.
Their studies, however, are exclusively in support of airborne laser bathymetry (“Airborne Laser Hydrography—System Design and Performance Factors”, NOAA Professional Paper Series, U.S. Department of Commerce, National Ocean Service 1, 1985). Guenther and his team have produced a large body of data and algorithms for processing such data.
Limitations due to fan-beam properties: Yet another obstacle to optimum practice of the innovations set forth in the U.S. Pat. No. 5,467,122 patent is the difficulty of obtaining uniform energy distribution and consistent divergence angle in the fan beam. Typically a single cylindrical lens is used to expand a laser beam of generally circular cross section, in just one dimension, into a fan shape.
In practice a high-energy pulsed laser beam is neither stable in size and position nor uniform in energy distribution, across the cross-section of the beam. Even a stable laser beam of uniform energy distribution, however, when thus spread to form a fan-shaped beam is nonuniform in intensity when it reaches the water surface (or any object plane)—due to long propagation distances required to reach the water at the extreme ends of the “fan”.
The propagation distance at each end of the fan is greater than that at the center by a factor equal to the secant of the fan half-angle. The beam divergence over this greater propagation distance proportionately reduces the beam brightness at the water surface—and the return reflection must also travel farther, additionally aggravating the brightness reduction at the detector.
It may be mentioned that the added travel distance also increases the return time at the fan-beam extremes. This delay, however, is wholly geometrical and therefore readily compensated in software; the accompanying brightness reduction cannot be resolved so easily.
Depending upon the character of the reflection process itself, the added return distance may produce either another factor of the secant, or instead another factor of the square of the secant. For components of the return beam that are generated through essentially specular reflection —such as the glints mentioned earlier, or even some specular portions of the volume backscatter—the incident beam angles in all directions should be approximately maintained in the return light and this implies that the same. proportional decrease in energy should develop again.
For components of the return beam that arise through true volume scattering, however, the distribution of energy in the reflection process should be omnidirectional. If it were distributed equally in all directions (not usual, but only a limiting case that can help to understand likely actual behavior), then the spatial distribution would follow a familiar inverse-square law—leading to attenuation of light from the fan-beam extremes by the square of the secant.
In practice a rather complicated and unknown added attenuation is likely to occur in those regions. The mix of phenomena can be mathematically modeled, and also measured empirically for a variety of conditions, to determine what factor of either added gain or added brightness would equalize volume backscatter under representative conditions. Application of that factor in increased gain or brightness may be expected to overcompensate with respect to brightness of glints—but this is unavoidable in view of the different reflection mechanisms involved, as explained above.
Because a gain-control approach would fail to equalize SNR across the fan-beam track, however, such an approach—although within the scope of the invention—is unappealing. It would lead to a systemic variation in SNR variation within every image, always.
In other words, information at the wings of the data array would be chronically both less clear and less reliable than information at the center. Hence the conceptual approach of adjusting the outgoing energy distribution in the LIDAR excitation beam is greatly preferable to a gaincontrol approach.
The '122 patent adopts precisely such an approach; it describes a way of roughly equalizing the energy received from the ends of the fan with that at the center. The rough equalization is obtained by halving a particular type of lens—and then reassembling the halves in opposite orientation.
This is done in such a way that rays are more concentrated at the limbs of the fan, tending to compensate very roughly for the longer divergence paths in those regions. Although extremely helpful, this system does not truly flatten the energy distribution along the intersection of the fan beam with the water surface, even in theory—and even for the volume backscatter as distinguished from the glints.
When the above-mentioned beam instabilities and nonuniformities are taken into consideration, the problem is far more severe. A laser beam generally varies in beam position, as well as energy distribution, from pulse to pulse.
The positional wandering takes the center of the beam off the center of the reassembled double-half-lens structure described above. This drift degrades the operating assumptions behind that device, and correspondingly disrupts its performance in equalizing energy at the limbs vs. the center of the fan.
The distributional drift enormously complicates any effort to systematically compensate for known departures from often-assumed “top-hat” or Gaussian energy distributions in the beam cross-section. Trying to correct for a constantly changing, unknown, high-power energy profile that is gone a nanosecond after it starts is a virtual impossibility with present-day technology—and stabilization of a high-power laser against both positional and distributional drifts is essentially prohibitive.
In fact even nominal alignment is a relatively onerous task. Preferably not simply the geometrical center of the beam but rather the effective center, in terms of maximum energy flux (or in terms of optimized energy flux over the entire fan-beam span) should be centered upon the reassembled lens structure.
Thus alignment becomes a matter of attempting to place the drifting effective center—of a beam of inhomogeneous and varying energy distribution, and varying position too —at the centerline of the reassembled lens structure. This is challenging.
As can now be seen, the related art remains subject to significant problems. The efforts outlined above—while praiseworthy—have left room for considerable refinement.